Electronic devices are increasingly able to provide various types of information and perform various tasks. For example, a user wanting to obtain information about a book can capture an image of the cover of the book and upload that image to a book identification service for analysis. In many cases, the user must actively focus a camera of the device on the object of interest and make that object occupy a majority of the image. Such an approach requires a significant amount of effort on the part of the user, and in some cases may not be possible given the location of the object. In other cases, an application executing on a device can attempt to recognize all the objects in an image, but such an approach takes a significant amount of processing and power, which can be particularly undesirable for small portable devices with limited battery life.